class: center, middle, inverse, title-slide # Sophisticated voters reward pro-climate parties and candidates in the wake of extreme weather events ## EPSA 2022<br><a href="https://soerendamsbo.github.io/misc-slides/epsa2022.html">soerendamsbo.github.io/<br>misc-slides/epsa2022.html</a> ### Søren Damsbo-Svendsen<br><a href="https://soerendamsbo.github.io">soerendamsbo.github.io</a><br>PhD fellow ### Department of Political Science<br>University of Copenhagen ### June 2022 --- background-image: url(data:image/png;base64,#media/bodil.png) background-size: cover .right[.grey[Image by Jeanne and John Bollerup-Jensen]] --- class: middle <img src="data:image/png;base64,#media/flooding_map_total.png" width="45%" style="display: block; margin: auto;" /> --- # A rarely harmful extreme weather event in Denmark - 2013 __storm surge__ after the winter storm 'Bodil' -- - Flooding measured with detailed, case-level **storm surge insurance data** (Danish Storm Council) -- - Aggregated at level of __1,386 polling districts__ -- - 3 treatment definitions (dummies): 1. **Rehousing:** 1+ cases of rehoused residents 2. **Damage:** storm surge damage (DKK) above average 3. **Flooding:** 1+ storm surge cases -- - Flooding matched with **main outcome:** _district vote share for pro-climate parties_ (DK, 1994-2019) -- - ... and, later, the __secondary outcome__: _election chances for pro-climate candidates_ -- -- - Difference-in-differences (DID) design -- *** > __RQ: How does an extremely costly flood event affect voting for pro-climate parties and candidates?__ ??? <img src="data:image/png;base64,#media/proclimate_candidates_kv2017.png" width="60%" style="display: block; margin: auto;" /> --- # Theory - Ample evidence that __extreme weather experiences can affect climate change perceptions and attitudes__ -- - But limited evidence that weather experiences affect __actual behavior__ &mdash; environmental or political -- - Several studies exist, but they are framed quite differently: -- - __natural disasters &rarr; retrospective voting__ - i.e., rewarding or punishing incumbents based on how they handled the disaster (or blindly) - __still mixed results:__ sometimes punishment, sometimes reward -- - How about **policy?** - more sophisticated voter behavior: updated __voter preferences__ on climate adaptation and mitigation - and/or policy or strategy shifts among parties and candidates - e.g., "a sizeable effect for pro-climate voting after experiencing a flood" (Baccini & Leemann, 2021) ??? Baccini, L., & Leemann, L. (2021). Do natural disasters help the environment? How voters respond and what that means. Political Science Research and Methods, 9(3), 468–484. https://doi.org/10.1017/psrm.2020.25 --- class: middle # Part I: support for pro-climate parties in parliamentary elections --- # Parallel trends: outcome evolution <img src="data:image/png;base64,#media/parallel_trends_plot_main.png" width="90%" style="display: block; margin: auto;" /> ??? <img src="data:image/png;base64,#media/parallel_trends_plot_all.png" width="3600" style="display: block; margin: auto;" /> --- # Parallel trends: pre-treatment balance <img src="data:image/png;base64,#media/balance_density.png" width="50%" style="display: block; margin: auto;" /> --- # Results: DID estimates <img src="data:image/png;base64,#media/did_estimates_2x2.png" width="3200" style="display: block; margin: auto;" /> ??? <img src="data:image/png;base64,#media/did_estimates_intensity.png" width="80%" style="display: block; margin: auto;" /> --- # Results: event study plots (placebo) <img src="data:image/png;base64,#media/did_estimates_eventstudy.png" width="85%" style="display: block; margin: auto;" /> --- class: middle # Part II: support for pro-climate candidates in local elections --- # Part II: support for pro-climate candidates Data: comprehensive __candidate survey__ of 2017 local election candidates (N=9,554) - w/ policy responses from 73% of candidates -- *Key variables:* - (1) *Outcome:* **being elected (binary)** in 2017, 2013 (and 2009) -- - (2) Running in a **treated municipality (binary)** w/ 10+ rehousing cases -- - (3) Running on a **pro-climate platform (binary)** w/ climate as key issue or high climate score (~32%) *** -- > __RQ: Are pro-climate candidates rewarded more (punished less) by flooding in affected areas?__ -- <br> _Triple differences (DIDID):_<br> .center[&rarr; we should expect **a positive difference in DID estimates** for pro-climate vs. non-climate candidates] ??? <img src="data:image/png;base64,#media/proclimate_candidates_kv2017.png" width="60%" style="display: block; margin: auto;" /> --- # Results: DIDID estimate <img src="data:image/png;base64,#media/did_estimates_2x2_candidates_table.png" width="75%" style="display: block; margin: auto;" /> ??? - `\(Treatment\:group \times Post\)` (-0.048\*\*\*): electoral punishment of -4.8 percentage points for non-climate candidates because of the flood (small reward for pro-climate) - `\(Climate \:candidate \times Post\)` (-0.008): no change over time in election chances of *non-climate candidates in non-flooded areas* - `\(Treatment\:group \times Climate \:candidate\)` (0.017): election chances of pro-climate candidates in flooded areas were the same as other candidates' *before the flood* --- # Results: DIDID estimate <br> .pull-left[ <img src="data:image/png;base64,#media/did_estimates_2x2_candidates_table.png" width="75%" style="display: block; margin: auto;" /> ] .pull-right[ - &rarr; approx. __6 percentage points__ difference (DIDID) - __pro-climate candidates are rewarded in post-flooding elections__ (long after the event) - not clear if voters change preferences or candidates (parties) change policy or strategy ] --- # Results: DIDID placebo <img src="data:image/png;base64,#media/did_estimates_2x2_candidates.png" width="75%" style="display: block; margin: auto;" /> <br> - also no effect (DIDID) for other candidate features, e.g., __incumbency__ and __pro-welfare platform__ --- # Conclusion > __RQ: How does an extremely costly flood event affect voting for pro-climate parties and candidates?__ -- - __Case:__ 2013 storm surge and surrounding Danish elections -- - __Data:__ Detailed insurance data, comprehensive candidate survey, administrative party vote shares -- - __DID design__ -- - __Results:__ - Pro-climate __parties__ (left-green) gain at least 1.5 pp - Pro-climate __candidates__ see their election chances increase by 6 pp vs. non-climate candidates -- - Importance of __policy__ and the __long-lasting effects__ suggest quite __sophisticated voting behavior__ (not *only* retrospective voting) -- <br><br> Still plenty of black boxes, analysis and interpretation to do... --- class: title-slide, middle, center # Thank you! &rarr; sdas@ifs.ku.dk <img src="data:image/png;base64,#media/banksy.jpg" width="45%" style="display: block; margin: auto;" /> .grey[Banksy (image by Zak Hussein<br>/ PA Images via Getty Images)]